The latest Survey USA poll of Virginia is getting a bit of attention today, as it shows Barack Obama to still be statistically tied with John McCain in the state – McCain 48, Obama 47, as befits a swing state. The full poll is here. And here's a little analysis of the situation.When you look closer, this poll has McCain losing southeast Virginia 53-40. As anyone who's followed Virginia knows, Navy absentees in the southeastern region of the state always give Republicans a slight bump over their polling in the area. And they're likely to be even more pro-McCain in this cycle, above and beyond their pro-GOP leanings.
SurveyUSA actually experienced the bump from Navy country in 2004 – they had Sen. John Francois Kerry in a similar statistical tie with George W. Bush all the way to the end, when Bush handily beat him by nine points. The PDF from November 2004 is here.
The eventual 2004 results are notable: yes, Kerry won Hampton, Norfolk, Surry, and Sussex – yet Bush's 33,000+ margin in Virginia Beach was more than Kerry's margin in all those counties combined, and his high turnout in other parts of the state overwhelmed Kerry's strong wins in Newport News and populous and wealthy Fairfax. For all the talk, it turned out to be an easy 54-45 win.
If history bears out, I think McCain is likely to take Virginia's southeast in November – but even a coinflip margin would likely translate to a statewide win for him. This, as in all other estimates about state results, is predicated on the idea that turnout trends aren't upset by a dramatic uptick in unexpected turnout from previously non-likely voters (in Virginia, this would come through a combination of Obama maximizing turnout in university communities like Williamsburg and black voters in Richmond, Hampton Roads, and Northern Virginia).
In other words: if Obama wins in Virginia, it will likely be part of a tide of victories nationwide, not the sole state deciding the result. McCain should hope that this state is a swing state in November, because the ground favors him.
Thanks to Michael Barone's analysis, we all know by now that Obama does very poorly in Appalachian areas – as he frames it, an "Academics versus Jacksonians" divide that Obama has yet to overcome. If this holds true in Virginia, it's likely that McCain will match or even exceed Bush's 2004 numbers in the southwestern portion of the state, and in the Shenandoah.
In the previous Survey USA poll of Virginia at the end of June, in which Sen. Obama lead by a statistically insignificant 2 points, the Democrat won the southeastern region by a 51-47 a four point margin – a far cry from the 13 point margin in this latest poll. And that's almost the entire basis for Obama keeping this state within the 3.9% Margin of Error – even if this poll is accurate, Obama has only improved on Kerry's 2004 margin by one percentage point in NOVA.
Did Obama really gain a nine point cushion in southeast Virginia in July? Or is this an outlier? And how good of a signal is it for the GOP that even – even with a poll that shows Democrat Mark Warner crushing the GOP's Jim Gilmore in the Senate race, even in a state bombarded by Obama's ads, and even with eight years of Democrats in the Governors mansion – with that nine point cushion, McCain still carries the state?
We shall have to wait and see.